Support pre-quantized models in aimet pass

[michaelgtuttle]

Describe your changes

Support accepting pre-quantized onnx models in aimet quantization pass.

Checklist before requesting a review

• Add unit tests for this change.
• Make sure all tests can pass.
• Update documents if necessary.
• Lint and apply fixes to your code by running lintrunner -a
• Is this a user-facing change? If yes, give a description of this change to be included in the release notes.
• Is this PR including examples changes? If yes, please remember to update example documentation in a follow-up PR.

(Optional) Issue link

[devang-ml]

Are there any expectations regarding how the model should be pre-quantized in this case?

[michaelgtuttle]

Are there any expectations regarding how the model should be pre-quantized in this case?

Good question. There are some limitations to what is accepted here, e.g., precisions that aimet does not support such as float8 would cause aimet to throw an error here.

For activation quantization it requires Q -> DQ pairs for quantized tensors, for weights it will accept either fp_initializer -> Q -> DQ -> Op or int_initializer -> DQ -> Op